A DRAM (Dynamic Random Access Memory) may lose data stored in memory cells as the time passes, unlike a SRAM (Static Random Access Memory) or flash memory. In order to prevent such a phenomenon, an operation of restoring data stored in memory cells at each predetermined period is required. Such an operation is referred to as refresh. The refresh is performed by activating word lines at least one time within a retention time of cells in a bank and sensing and amplifying data stored in the cells. The retention time refers to a time during which data stored in a cell may be maintained without refresh.
FIG. 1 is a diagram showing interference which occurs between word lines when the word lines are activated.
Referring to FIG. 1, a second word line WL<2> neighbors first and third word lines WL<1> and W<3>. In general, when word lines are activated, a high voltage VPP is applied to the word lines such that electromagnetic interference occurs between the word lines. When the first and third word lines WL<1> and W<3> are activated, a small amount of voltage is generated in the second word line WL<2> in response to electromagnetic interference between the first and third word lines WL<1> and W<3>. Therefore, a leakage current is generated through cell transistors that couple the second word line WL<2> to cells, respectively. The retention time of each cell is reduced by this leakage current.
Accordingly, when entering a refresh mode immediately after the first and third word lines WL<1> and W<3> adjacent to the second word line WL<2> are activated, a refresh fail may occur because the retention time of the cells coupled to the second word line WL<2> was reduced. That is, referring to FIG. 2, when a refresh pulse PSRF having a preset period t0 is outputted to perform a refresh operation from a time point T1 to a time point T2, the refresh pulse RSRF is generated after the retention time of the cells coupled to the second word line WL<2> passes. Therefore, a refresh fail occurs.